View Online / Journal Homepage / Table of Contents for this issue

[19471 Theory of Chromatography. Part I V . 1315

241. Theory ofChromatogruphy. Part I V . T h e Influence of

Incomplete Equilibrium o n the Front Boundary of Chrornatogram
and o n the Effectiveness of SepuTation.
By E. GLUECKAUFand J. I. COATES.
It is shown that in the chromatographic column, under the usual working conditions, com-
plete local equilibrium is not obtained. A theory is developed for the effect of incomplete
equilibrium on chromatographic boundaries, and the form of observed boundaries agrees
Published on 01 January 1947 on http://pubs.rsc.org | doi:10.1039/JR9470001315

with those calculated. On decrease of grain size and flow velocity, equilibrium becomes more
complete, and then some chromatographic separations can be effected with ease which are
impossibIe by the usual methods; c.g., the separation of nickel and cobalt ions on alumina.

(a) Chromatographic Boundaries.-The influence of incomplete local equilibrium is very

similar to a diffusion effect and so shows itself most markedly at the boundaries of the chrom-
Downloaded by University of Toronto on 19 November 2012

atograms where high concentration gradients exist. Normal binary chromatograms when fully
developed have, in the absence of disturbing effects, the form shown in Fig. 1 of the preceding
paper. We may distinguish here two types of boundaries : (i) sloping rear boundaries such as
A B , CD, and C‘D‘, and L’Q, and (ii) sharp front boundaries such as the threshold boundary
HG or PQ and the inter-boundary E F , E’F’ or LK, L’K’ which is the front of solute 2. As
has been shown by De Vault ( J . Amer. Clzem. SOL, 1943, 65, 532) the two front boundaries
have normally (i.e., for isotherms concave against the c-axis) self-sharpening properties. This
means that, even if by some artificial means HG were made sloping, it would again become
vertical during the process of development-at least theoretically. This self-sharpening tend-
ency increases with the curvature of the isotherm and it is nil for linear isotherms (see
De Vault, eqn. 7).
In the case of the threshold boundary HG, the self-sharpening effect is usually very con-
siderable, as linear single-solute isotherms are rare. I n the case of the inter-boundaries EF
the corresponding function depends essentially on the “ exchange ” isotherm between the
two solutes. If the two solutes are very different and easily separable, this boundary, too,
usually shows considerable self-sharpening, but if the two solutes are very similar, the exchange
isotherm is practically linear and then this effect is negligible. The inter-boundary is then
referred to as a non-sharpening front boundary.
(b) Attainment of Complete EquiZibrium.-In order to study the effect of non-equilibrium
and diffusion phenomena, chromatographic elution curves were measured, using adsorbents
of different grain size. The adsorbents were ground to a diameter as low as 0.01 mm. (as
compared with the commercial grain sizes of about 0.08 mm.), and a uniform grain size was
obtained by elutriation in air streams of varying velocity.
The flow rate of the solvent in the chromatographic column was varied from 0.001 cm./sec.,
which means 1 drop in 2 mins. for a tube of 1 cm. diameter, to the usual flow rates of 10-60
drops per min.
Systematic experiments showed that the quality and sharpness of non-sharpening inter-
boundaries, q., such as between copper and manganese ions on Zeo-karb H.I. or H.I.P.,
continuously increased with reduction of grain size and flow rate. There must be an optimum
flow rate, as the latter can obviously not be infinitely decreased without introducing disturb-
ances on account of diffusion. But this latter effect is so small compared with the non-
equilibrium effect and other disturbances that even at the lowest flow rates used an improve-
ment was still noticeable compared with faster flow rates.
Under the conditions mentioned above (d = 0.01 mm., F = 0.001 cm./sec.), we were able
to obtain, for a column of 1 m. length, non-sharpening inter-boundaries between 0.6 and 1 cm.,
and self-sharpening front boundaries extending on one occasion over only 0.10 cm., the former
corresponding to O-5-lyo, the latter to 0.1% of the threshold volume.
That these conditions are not obtained with the usual practice can be seen from Fig. 1Oa
of Weil-Malherbe (J., 1943, 303), which contains about the best chromatographic experiments
published. Here a non-sharpening front (benzpyrene on silica gel) extends over 25% of the
threshold volume, and a strongly self-sharpening one (benzpyrene on alumina) over 12% of
the threshold volume. A check on the form of non-sharpening front boundaries is therefore
essential in all experimental work connected with the study of elution curves and, in particulax,
in such cases where elution curves are used for the calculation of adsorption equilibria (see
Glueckauf, Nature, 1945, 156, 748, and Part 11).
 1316 Gheckauf and Coates : View Online

Although it is apparent that non-equilibrium phenomena can easily occur in the case of
exchange materials, it is not usually realised that the same, though perhaps to a lesser degree,
also applies to the so-called adsorbents " like alumina and calcium carbonate, as adsorption
I'

takes place not only on the surface, but also inside the grains. The reason for this is that
most experimenters investigating the form of bands have worked with single solutes which
usually have strongly self-sharpening fronts. These maintain a reasonable sharpness even if
considerably disturbed by non-equilibrium phenomena. Consequently, observers are easily
deceived by this appearance of moderately sharp fronts, which are usually taken as proof of
the absence of non-equilibrium phenomena.
How far the process of wishful thinking can go is shown, e g . , by Jacobs and Tompkins
Published on 01 January 1947 on http://pubs.rsc.org | doi:10.1039/JR9470001315

(Trans. Furaduy Suc., 1945, 41, 401, Fig. 1, Curve A), who convinced themselves-with the
help of the equilibrium curve applied to a diffuse non-equilibrium boundary-that, in spite
of the very diffuse eluted front which they found, the front of the adsorbate in the column

FIG. 1.
Downloaded by University of Toronto on 19 November 2012

curve b

4
b &5 io i5 zfl
C.C. o f e h a t e before and after centre of boundary.
Experimentally measured boundaries. (a) Self-sharpening threshold boundary of N-CoCl, solution i n
column of acid Zeo-karb. ( x - x - X ) (compare with curve calculated for bco = 10 in Fig. 4).
(b) Non-sharpening inter-boundary between frontal band of 58C0C12and mixed band of 59CoC12and
SOCoC1, in same column of Zeo-karb (0 - 0-0) (symmetrical S shape).

must have been very sharp. By doing so, they overlooked the fact that a sharp front of
adsorbate moving through the column must deliver the solute a t the end in an equally sharp-
fronted elution curve-simply by reason of the conservation of matter (for details see eqn. 5).
The non-equilibrium phenomena, however, become very obvious when fronts are con-
cerned which have little or no self-sharpening properties, e.g., the boundary of one solute
against another of similar adsorption affinity. As it is just such a boundary which occurs
a t the point of separation of two similar solutes (boundary E F in Fig. 1 of Part 111),these
non-equilibrium effects greatly reduce the efficiency of the separation.
(c) Form of the Disturbed Selfsharpening Front Boundary.-No theory has so far been given
for the theoretical form of a front boundary disturbed by non-equilibrium phenomena in the
case of a non-linear isotherm. Weiss (I., 1943, 301; eqns. 59-62 and Fig. 5) has produced
a schematical representation for the case of a linear isotherm which, however, apart from the
diffuseness, does not really represent the experimental facts (see our Fig. 1, curve b, showing
that the measured elution curve of a front boundary for a strictly linear isotherm has a sym-
metrical sigmoid form. A similar conclusion has also been reached by Klotz, Chem. Rev.,
1946, 39, 243). A satisfactory theory for the case of the linear isotherm has been produced
by Dr. H. London which, however, has not yet been published.
I n the case of a self-sharpening front boundary caused by a non-linear isotherm, the general
differential equations become very complicated, but, after the front has travelled some distance,
a final form is attained which can be represented by a simple equation. Here the effects of
non-equilibrium and self-sharpening just balance each other. For simplicity the calculations
are carried out for the '' undeveloped " chromatopam where the front has the original
concentration 60.
 [19471 Theory of Chromatography. Part IV. 1317
View Online

In the case of non-equilibrium, the usual chromatographic equation of mass conservation

(see Part 11, eqn. 2), viz.,
+'
[Av/Az]~ = df*(q/dc 61 . . . . . . . . (1)
changes to
[Av/Ax]~ = cc +
. . . . . . . . (2)
where q* is the amount actually adsorbed, the missing amount (f*(c) - q*) being due to dis-
equilibrium. When at a given point (x) of the front boundary the concentration rises during
the passage of the front through this point, q* is, with good approximation, represented by the
equation
dq*/dt = K(f*,,, - q * ) . . . . . . . . (3)
Published on 01 January 1947 on http://pubs.rsc.org | doi:10.1039/JR9470001315

which means that the rate of diffusion into the grains is essentially proportional to the amount
still required to produce equilibrium. Here K is a diffusion factor (the inverse of the time in
which a grain of adsorbent reaches l/e of its equilibrium adsorption), and K = where
D is the diffusion constant of the solute in the solid material, d is the grain diameter, and
Downloaded by University of Toronto on 19 November 2012

q(<1/2) depends on the geometry of the grains.

FIG.2.

(u-~ f ) K / 2 * 3 F
SeEf-sharpeningfront boundaries for solutes obeying Freundlich isotherms q = acP for different values of p.

After competition between the self-sharpening and the non-equilibrium effects has resulted
in a final form of the front boundary, the whole front boundary and thus also t h e point of
concentration G move with the usual velocity of the sharp undeveloped front boundary
[Av/Ax]~ = f*(p$/Co cc + . . . . . . . . (4)
It follows from equations (2) and (4) that in this final state
5 dq*/dC = f*(caJCO . . . . . . . . . (6)
This means that, no matter what the form of the equilibrium isotherm, q* must always be a
linear function of c. Eliminating q* by means of (5) in (3) and replacing t by v / F (where
F = flow velocity of solvent, in c.c./sec.), we get
dc/du = (K/F)[f*,,,. Co/f*ce, - C] . . . . . . . (6)
Equation (6) has a particularly simple form in the case of the Freundlich isotherm
f*(c)= A * CP . . . . . . . . . . (7)
Integration of (6) then leads to the equation for the elution curve of a disturbed self-sharpening
front (final state) :
-F
v-ver,=- . log [l - (c/cO)(1-p'] . . . . . . (8)
K(1 - P )
Similarly one obtains for the frontal distribution in the column
-F
- . log r i - ( q * / 4 0 ) ( 1 - ~ ) ~
GO
xf - x = . . . . .
K(1 - p ) $ 0 )
~

- (9)
49
 1318 Glueckauf and Coates : View Online

where xr is the most advanced point of the front boundary. Equation ( 8 ) requires that the
length of the eluted front is independent of the concentration co, which is perfectly shown in
Jacobs and Tompbins's Fig. 2 (Zoc. czt., p. 402). Equation (8) also shows the effect of p on the
slope and extension of the front boundary, which is demonstrated in the experiments of Weil-
Malherbe (Zoc. cit., Fig. lOa), though not quantitatively, since, with the short columns used,
the final state is not attained for values of 9 near unity.
Fig. 2 gives the theoretical curves of eluted front boundaries in the stationary state for
various values of +. It shows that for small values of the Freundiich exponent p (c.g., p = +)
the front rises sharply but reaches the maximum (GO) only asymptotically. For comparatively
large values of p it has a sigmoid form, which for p = -1 reaches both the values c = 0 and
Published on 01 January 1947 on http://pubs.rsc.org | doi:10.1039/JR9470001315

e = GO asymptotically.
FIG.3.
Downloaded by University of Toronto on 19 November 2012

Self-sharpening front boundaries for solutes obeying Langmuir isotherm q = ac/(l + bc) for
different values of bco.
Equation (8) also shows that the question of grain size and flow rate comes into play to a
very marked extent only when very similar substances have to be separated, where the inter-

- -
boundary is due t o an exchange isotherm which, for very similar substances, is almost linear
(1 - p = 0 ; thus v - vt = co ). No final state is attained by the front boundary when
p = 1 (linear isotherm), under which conditions there is no self-sharpening effect, and, con-
sequently, the slope of such a front boundary continues to decrease indefinitely.
Very similar conditions arise in the case of the Langmuir isotherm
f*(c)= ac/(l + bc) . . . .
Here integration of eqn. ( 6 ) leads to the equation for the elution curves :
K 1
+
vc - "c" = - T [ D .log (GO - c)/c
2 1.
log 2 (GO - c)/cO . . .
The form of the front boundary here depends on bcO which, according to the physical inter-
(11)

pretation of the Langmuir isotherm, is the ratio of covered to free adsorbing surface. As can
be seen from Fig. 3, the self-sharpening tendency increases with beo, ie., with increasing surface
saturation. A comparison with Fig. 2 shows that there is a complete parallelism in the form
of the front boundaries for the two isotherms, small values of 9 corresponding to large values
of bcO and vice veysa.
The diagrams of two accurately measured front boundaries are shown in Fig. 1. Curve a
is the eluted front boundary of a l ~ - [ C o C l ~band
] after passing 3g of acid permutite. "The
isotherm is approximately of the Langmuir type with a value of b 3-4 (thus bcO = 3 4 ) .
A comparison with the corresponding calculated curves in Fig. 3 shows a good agreement.
Curve b shows an inter-boundary between a mixture of WOand OOCo against a front-band of
pure W o , where the rise in the concentration of SOco has been measured. As the exchange
isotherm between the two isotopes is strictly linear, there is no self-sharpening to be expected,
and the experiment shows, in agreement with the theory, a symmetrical S-form for this front
boundary.
 [1947] Theory of Chromatografihy. Part I V . 1319
View Online

The elution curve of the boundary of the two isotopes was measured by means of the p
activity of 6oCo (T = 5.5 years) in single drops of eluate which were dried on filter-paper.
(d) Eflect of Non-equizibrium on Seflarafion of Two Solutes.-A typical example of the
importance of such considerations is shown in the separation of copper and nickel on com-
mercial alumina by Jacob and Tompkins (Eoc. cit., p. 398, Fig. 3), using a linear flow velocity
of 0.07 cm./sec. Even after the band has travelled several times the distance which they
expect to be required for complete separation, the mixed part of the band still contains about
70% of the nickel and about 30% of the total copper.
We obtained very similar results with copper and.manganese (the latter being used instead
of nickel owing t o its easy volumetric determination). Using a flow velocity of 0.027 cm./sec.
Published on 01 January 1947 on http://pubs.rsc.org | doi:10.1039/JR9470001315

and 0-26~-sulphuric acid for elution on 2.2 g . of commercial alumina, we obtained the elution
curve of Fig. 4a, which shows the retention of a mixed band with 47% of the manganese and
about 25% of the copper. (Figs. 4a and b were obtained by analysing single drops of eluate.)

FIG.4.
Downloaded by University of Toronto on 19 November 2012

0-25[1
020 _----*-------*

0 1 2 3 4 5
voz, C.C.
0-25
0.20
0-15
Lj
0 70
0-05
0
0 7 2 3 4 5
VOL,c.c
Elution curves of mixture of CuSO, and MnSO, chromatographed o n 2.2 g. of Also,. Devekofimenf with
O*%N-H,SO,.
Abscissa : volume of eluate after appearance of solute in eluate.
Ordinate : concentration of solutes in eluate (milliequiv.1c.c.).
(a) Diameter of Al,O, = 0.08 mm. ;*$ow velocity = 0.027 cm./sec.
(b) Diameter of Al,O, = 0.01 mm. ; $ow velocity = 0.0014 cm./sec.

By reducing the grain size from 0.08 to 0.01 mm. diameter and the flow velocity to 0.0014
cm./sec., and otherwise using identical conditions, the effect of the diffusion and non-equilibrium
phenomena was much reduced (see Fig. 4b). The mixed part of the band now contained only
20% of the manganese and 14% of the eluted copper, and in some experiments still less.
(Actually, separation appeared to be much more complete in the column, and the small amount
of mixing observed in the eluate seemed to be due to the mixing of 2 or 3 drops at the bottom
of the tube.) Also, in other respects, the two experiments show marked differences. In the
second experiment (Fig. 4b) all the front boundaries of the Mn++, Cut+, and H+ are fairly
steep, and the flat levels of the manganese and copper bands show clearly that these bands
are not being “ developed ” by a diffuse rear boundary, a feature which is lost completely
through non-equilibrium phenomena in Fig. 4a, and in the copper elution curves of Jacob and
Tompkins (Zoc. czt., p . 402, Fig. 2).
Although, under ordinary working conditions (commercial grain size and usual flow rate),
some separation is still shown in cases where the differences in adsorption affinity and the
self-sharpening tendencies of the inter-front are comparatively large, as is the case for Cu-Ni
and Cu-Mn, it becomes impossible to separate solutes of very similar adsorption affinity, which
consequently have fairly linear qchange isotherms, and whose inter-front is practically non-
self-sharpening.
 1320 Theory of Chromatography. Part I V . View Online

Such solutes as, e.g., the group Ni, Co, Cd, FeII, are usually considered as non-separable
(see Schwab and Jockers, 2.angew. Ghem., 1937, 50, 646), but their inseparability is entirely
due to the disturbing influence of diffusion and non-equilibrium phenomena which almost
completely undo any chromatographic separation. No difficulty is experienced in effecting
an almost complete separation-with adjoining bands, of course-between any two of these
solutes if a smaller grain size and a lower rate of flow is employed. To give an example :
+
0.15 C.C. of O-~N-N~SO,O-~N-COSO,were developed with 0*25~-H,sO,. The original band,
in a column of alumina of 8 mm. diameter, had a length of 4.6 mm. The band at first broadened
on development with the more dilute acid and, by the time the just visible rear band of nickel
had travelled 6 mm., separation was complete, the nickel occupying a position between 6 and
Published on 01 January 1947 on http://pubs.rsc.org | doi:10.1039/JR9470001315

12 mm., and the cobalt that from 12 t o 18 mm. Treatment with dimethylglyoxime and
ammonia showed that, if there were any overlap between the nickel and the cobalt zone, it
must have been less than 1 mm. A quantity of 0.3 g. of alumina thus proved t o be sufficient
to separate almost quantitatively 0-15 milliequiv. of the inseparable ” cobalt-nickel sulphate
‘I

mixture.
Downloaded by University of Toronto on 19 November 2012

It should also be mentioned that, as a result of using finer grain, the quality of packing was
much improved. The boundaries obtained were absolutely straight. Fringes-only observ-
able under the microscope-did not exceed 3-4 grain diameters, i.e., 0.04 mm. on any of
the boundaries of Fig. 4b.
FIG.5.

Chromatogram of mixture of CuSO, and CoSO, on A1,0,. Develoflment r i t h MnSO, solution. C u band
remains at the top of the column. Three stages in the “ development of the Co band are shown.

(e) Separation into Isolated Bands.-As the slow separation of metal ions on fine-grain
alumina proved so successful, an attempt was made to obtain a complete separation into
isolated bands. With acid as developing agent, this is of course not possible, as hydrogen
ions are the most strongly adsorbed ions and, consequently, the bands cannot be expected
to separate (see Glueckauf, Part I, P ~ GRoy.. SOC.,1946, A , 186, 50, case 4). I f a complete
separation, e.g., of copper and cobalt bands is required, development must be attempted with
a less adsorbed solute, such as Mg++ or Mn++solutions.
Fig. 5 shows the approximate distributions of the separated bands observed during a
+
chromatographic separation of 0-15 C.C. of O-~N-CUSO, O-~N-COSO,,developed with 1, 1.5,
and 4.5 C.C. of O.~N-M~SO, solution. Already after 1 C.C. of developing solvent had been
added, the separation of copper and cobalt seemed almost complete, resulting in adjoining
bands of about equal width. Further development hardly altered the position of the copper
band. The front of the cobalt band continued to move down the column, but the tail end,
though greatly decreasing in concentration, never moved away from the front of the copper band,
a typical sign that cobalt ions follow a Freundlich isotherm even in the presence of 0*5~-Mn++.
(During the development it looked as if the bands were actually completely apart, but
subsequent treatment with ammonium sulphide showed that the cobalt band reached right
back to the copper front.)
Thus, if separation into two entirely separated bands were required, it would in this case
be necessary either to develop with a solute of intermediate adsorption affinity (e.g., zinc),
or to add such a solute to the mixture and develop with acid.
Somewhat similar is the separation of nickel and cobalt with O-~N-M~SO, solution as
developing agent. Here both bands attained a triangular shape of approximately equal
length, the nickel band beginning a t the top of the column, and the cobalt band at the fairly
sharp front of the nickel band. (The cobalt band was here not actually a triangle, but began
 [1947] Theory of Chomatography. Part V . 1321
View Online

with a definite concentration, as must be expected for Freundlich isotherms. This is shown
in detail in Figs. 4, 6 , and 6 of Part V.)

Our thanks are due to Dr. €3. London, University of Birmingham, who first drew our attention to
the great importance of non-equilibrium phenomena for the separation of solutes of very similar adsorp-
tion aflinity.
DURHAM IN THE UNIVERSITY
COLLEGES OF DURHAM. [Received, November 26t12, 1946.1
Published on 01 January 1947 on http://pubs.rsc.org | doi:10.1039/JR9470001315
Downloaded by University of Toronto on 19 November 2012